An Enhanced Multi Featured Video Object Detection and Tracking Using Threshold Filters

摘要

Video object segmentation is a primary task for various computer vision applications in video surveillance. Recently, many research works has been designed for video object segmentation and detection. However, video object segmentation based on multi features like size, color, texture, shape, intensity was not efficiently performed. In order to overcome this limitation, an Enhanced Multi Featured based Video object Detection (EMFVD) method is developed in this paper. The main goal of EMFVD method is to segment the video object based their features like size, color, texture, shape, intensity and to efficiently detect the moving object in video sequence using threshold filtering technique. Initially, EMFVD method takes the video file as input and then performs the video object segmentation task with the assist of filtering technique. During the segmentation task, video frames are segmented based on size, color, texture, shape, intensity and noises in video frames are removed which results in improved image quality. After performing the object segmentation, EMFVD method tracks the moving video objects in video sequence with the help of filtering technique. Finally, EMFVD method performs efficient moving object detection by applying the threshold technique which results in enhanced video object detection accuracy. In EMFVD method, threshold technique is done with the assist of Gaussian-based Neighbourhood Intensity Proportion (GNIP). Proposed EMFVD method used video images obtained from Internet Archive 501(c) (3) for conducting experiment work. Experimental analysis demonstrates that the EMFVD method is able to improve the video object segmentation accuracy by 13% and to reduce mean square error rate by 30% when compared to the state-of-the-art works.